Anthropic launches Claude Fable 5.1 and Mythos 5.1 with cheaper pricing, new data controls

Fable 5.1 is generally available through the Claude API, Amazon Bedrock, Google Cloud, and Microsoft Foundry, while Mythos 5.1 is restricted to vetted organizations in Anthropic's trusted access programs for cybersecurity and life-sciences research. The two models share the same underlying architecture but differ in safeguard layers, with Mythos designed to support sensitive work such as vulnerability discovery and exploit research.

Anthropic says Fable 5.1 sets new records on several benchmarks. On Terminal-Bench-Science 0.1, an agentic scientific research benchmark, Fable 5.1 scored 52.6%, more than double the 24.7% achieved by the previous Fable 5. On Terminal-Bench 4.0, a CLI coding benchmark, Fable 5.1 reached 55.8% and Mythos 5.1 reached 60.9%. The company also reported a 60.9% score on Humanity's Last Exam without tools, rising to 65.0% with tools.

The biggest commercial change is a 75% reduction in cache-read pricing, from $1.00 to $0.25 per million tokens. Anthropic says this lowers the cost of running typical workloads by about 25% and highly agentic workloads by up to 45%, while base input and output prices remain unchanged at $10 and $50 per million tokens. The company expects the lower cost to make long-running agents more economical.

Anthropic also walked back a controversial data-retention policy introduced in June with the Fable 5 launch, which required 30-day retention for all traffic on its most advanced models. After feedback from business customers, the company introduced Enterprise Frontier Safeguards, a system that lets organizations control how their data is reviewed, stored, and managed, with automated safety monitoring that requires no Anthropic human review. Data will be stored on customers' cloud servers rather than Anthropic's own infrastructure. The service will be free. According to TechCrunch, Enterprise Frontier Safeguards will roll out to users in June, while Anthropic said it will roll out in phases with broader availability targeted for this fall. The company also stated it has never trained on enterprise data without permission.

The new models feature more relaxed guardrails compared to their predecessors. Fable 5.1 can now be used to identify software vulnerabilities, though more advanced tasks such as exploit generation and binary-based vulnerability scanning are still redirected to Opus models. Anthropic says biology-related safeguards now fire 85% less often on benign requests. Mythos 5.1, which has fewer safety restrictions in cybersecurity and biology, is described as possessing the strongest cyber capabilities of any model the company has released.

The launch comes a day after Anthropic signed a $35 billion infrastructure deal with cloud startup Lambda, according to sources cited by the Wall Street Journal. The capacity is expected to be housed in a Hut 8 data center in Nueces County, Texas, with Nvidia holding the lease. Earlier last week, Anthropic inked an even larger contract with Nscale Global Holdings, reportedly valued at $45 billion, to rent about 460 megawatts of computing capacity in a West Virginia data center. These deals signal Anthropic's expectation that its infrastructure needs will grow significantly as it races to compete with rivals like OpenAI.

Anthropic generates most of its revenue from enterprise sales, and the company is widely expected to pursue a major IPO. The new pricing and privacy controls are aimed at easing concerns from large customers who had objected to the previous data-retention requirements.